Summary The pathways involved in embryonic development have been a rich resource for understanding disease in adults, as well as being critically important in tracing the effects of genetic lesions and environmental poisons in the fetus. Frog embryos have been particularly useful due to the large size of the frog egg and embryo. New tools we developed for measuring the expression of RNA at a single-cell level, and advances in protein and phosphopeptide measurement technologies, offer hope for dramatic progress in understanding how signals involved in the maturation of the embryo direct individual cells to adopt specific fates. Our first goal is to define cell types using single-cell transcriptomics, and to define the lineages that result in specific cell types using high resolution temporal mappings. Targeted transcriptomics and proteomics of important molecules involved in specifying cell fate, such as transcription factors, will provide an index of the levels of signaling activity in each individual cell. This will result in an unprecedentedly detailed molecular picture of the factors involved in producing the phenotypes, and their interconversions from the early cleavage stage to the middle of organogenesis. The Xenopus model system allows us to dissect out portions of the early embryo that differentiate to ectoderm if not disturbed, called the animal cap. In the context of the embryo the cells in the animal cap receive a number of developmental signals, including Nodal, BMP, and Wnt. Combinations of these three signals (in different proportions) are capable of generating many of the major tissues. We will expose animal caps to a matrix of these three signals and trace the differentiation pathways that result, using single-cell RNA sequencing. This study of the molecular roots of differentiation decisions will be used to develop a mathematical approach, based on machine learning, to predicting the results of an attempted perturbation of the development of Xenopus. We will ask whether cell types are carefully specified by tightly controlled combinations of ligands or whether there are default states that are hard to escape from (basins of attraction), that therefore form the majority of embryonic cell types. The answer to this question is central to our understanding of how the Xenopus embryo reliably develops into a frog, and will accelerate efforts to create computational methods to predict the behavior of other biological pathways such as those involved in cancer.